CN101879073B - Ultrasound diagnosis apparatus - Google Patents

Abstract

The present invention provides an ultrasound diagnosis apparatus, comprising an image generator configured to execute transmission/reception of ultrasound waves to chronologically generate ultrasound image data of plural frames; a multiresolution decomposition part configured to hierarchically perform multiresolution decomposition on the ultrasound image data to acquire first-order to nth-order (n represents a natural number of 2 or more) low-band decomposition image data and first-order to nth-order high-band decomposition image data; a feature amount calculator configured to calculate a feature amount based on the acquired low-band decomposition image data; a filtering processor configured to perform a filtering operation on the calculated feature amount; and a multiresolution composition part configured to execute multiresolution composition using the low-band decomposition image data and high-band decomposition image data to generate a composite image. Thus, the apparatus can efficiently reduce change in speckle/noise in the temporal direction and perform a process without a time phase delay.

Description

Diagnostic ultrasound equipment

Technical field

The present invention relates to subject transmission ultrasound wave and based on the diagnostic ultrasound equipment that obtains ultrasonography from the echo of subject.

In more detail about the ultrasonography that generates being carried out the diagnostic ultrasound equipment of treatment for correcting.

Background technology

The ultrasound wave that diagnostic ultrasound equipment will produce from the oscillator that is built in ultrasound probe sends to subject.Diagnostic ultrasound equipment receives the hyperacoustic echo from the subject tissue that is sent through above-mentioned oscillator, and generates, shows the ultrasonic tomogram image of subject according to the echo that receives.

But, different from the reception signal phase place separately of a plurality of approaching subject tissues, therefore can disturb.Because this disturbs, under the situation of net amplitude only, cause visual difference, generate mosasic picture pattern.This picture pattern is called speckle (speckle).This speckle counteracts the position and the shape on the border of observing the subject tissue exactly.

And, because generating noise also hinders observation accurately.This speckle or noise not only produce at direction in space, also on time orientation, produce.

In the past, the various processing methods of speckle and noise have been proposed to be used to reduce.

IIR (the Infinite Impulse Response: the technology of wave filter infinite-duration impulse response) of direction service time for example, has been proposed.This technology is such technology: be I (t) when being located at the image that time t obtains, to handle the image that obtains through wave filter be J _tThe time, use and satisfy J _t=(1-a) I _t+ aJ _T-1The wave filter of the relation of (wherein, a≤1).Through this technology, can reduce the speckle and the noise of change in time.

But this technology is passed through in current data I _tThe ratio of (1-a) add past data J _T-1The ratio of a, thereby reduce noise.Therefore, under the situation that be organized as object violent, in eclipsed image, move violent separately the image organized as after image and residual, have the problems such as fuzzy that produce with actions such as diaphragm and hearts.

In addition, be that each frame generates under the situation of image, according to the motion vector of each image in the frame in past (the for example image of internal organs) detected image to making the unit of a tension fault as the set of required data.

And then proposing has such technology: use the position of the image in this motion-vector prediction next frame, Filtering Processing etc. is implemented in the position of this prediction.

Should technology if use, then can comprise the Flame Image Process of the offset of the image that produces because of effluxion.

But, as stated,, noise also need detect the processing of motion vector except reducing processing, therefore there is the problem slowly of handling.

And, in the past,, used speckle/noise reduction filter for clear and definite tissue part.

As described speckle/noise reduction filter, use the wave filter that has utilized the direction characteristic amount in two dimension or the three-dimensional space.So-called direction characteristic amount be meant the edge size or edge towards.Here, the edge is meant the part that the brightness of that kind such as the outside of tissue significantly changes and refers to the part beyond the noise.

And then, as the above-mentioned wave filter that utilizes the direction characteristic amount, can use on the image direction on the edge of fuzzy wave filter or with the wave filter of the contrast of on orthogonal direction, strengthening image at edge etc.Get the wave filter etc. that point arranged side by side is obtained meansigma methods as making on the image direction on the edge of fuzzy wave filter, for example can using.

And, as with the wave filter of on orthogonal direction, strengthening contrast at edge, for example can use according to threshold value to make the brightness of the Outboard Sections at edge reduce, make wave filter that the brightness of the inside part at edge improves etc.

Make image blurring wave filter through using on the direction on the edge of, thereby can access deep or light level and smooth image on the direction on the edge of.

And, through use with the edge on orthogonal direction, strengthen the wave filter of the contrast of image, thereby obtain the image of the clear and definite and accentuate tissue part in edge.

And then, proposed to use multiresolution to decompose as the technology of obtaining the method for above-mentioned direction characteristic amount.

So-called multiresolution decomposes and is meant, signal of video signal is decomposed into the have different frequency bands method of a plurality of images of (spatial frequency band) arbitrarily.

Decomposition can be used the Wablet conversion as multiresolution.This Wablet conversion is decomposed into 2 of input image ^-1The image of the low-frequency component of size and the image of radio-frequency component.

Based on the image that has carried out after this Wablet conversion, use diffusion filter to detect the edge of image of spatial frequency as low-frequency band, and to each pixel obtain the edge towards.

And then; Proposed following technology: tangential direction on the edge of makes image blurring wave filter carry out the smoothing of pixel, and normal direction on the edge of uses the wave filter of strengthening contrast to carry out the sharpening (for example with reference to TOHKEMY 2006-116307 communique) of pixel.

The tissue part of image that thus, can clear and definite low-frequency component.

But multiresolution that TOHKEMY 2006-116307 communique is put down in writing decomposes and is to carry out noise about direction in space to reduce and handle based on the Flame Image Process of each wave filter.

But, in the technology that TOHKEMY 2006-116307 communique is put down in writing,, do not carry out that this multiresolution decomposes and based on the Flame Image Process of each wave filter for the image that comprises time orientation.

Therefore, reduce in the method for handling, in respect to the image of time orientation, can produce noise and speckle at this noise that only carries out about direction in space.

And, when direct application is directed against spatial Filtering Processing, meeting generation time Phase delay, existence can't show the problem of the image (real time imaging) that has just generated.

Summary of the invention

The present invention is directed to the coordinate (time+space) that comprises time orientation, use the utilization carried out at direction in space the noise of direction characteristic amount reduce and handle.

Especially, the object of the present invention is to provide a kind of such technology: through use not can the generation time Phase delay processing, thereby in the ultrasonography of the violent tissue of relevant action, the noise and the speckle of reduction time orientation, accentuate tissue part.

The diagnostic ultrasound equipment of first mode of the present invention is characterised in that this diagnostic ultrasound equipment possesses: image production part spare, and through the execution ultrasonic transmission/reception, thereby to generate the ultrasonography data of a plurality of frames by the seasonal effect in time series mode; The multiresolution Knock-Down Component carries out multiresolution by level to above-mentioned ultrasonography data and decomposes, and obtains the 1st time～the n time low-frequency band decomposition view data and the 1st time～the n time high frequency band and decomposes view data, and wherein, n is the natural number more than 2; Characteristic quantity is calculated parts, decomposes view data according to the above-mentioned low-frequency band that obtains and calculates characteristic quantity; The Filtering Processing parts are implemented filtering operation to the above-mentioned characteristic quantity of calculating; And the multiresolution compound component, it is synthetic to use above-mentioned low-frequency band decomposition view data and above-mentioned high frequency band decomposition view data to carry out multiresolution, thereby generates composograph.

The diagnostic ultrasound equipment of second mode of the present invention is characterised in that this diagnostic ultrasound equipment possesses: image production part spare, and through the execution ultrasonic transmission/reception, thereby to generate the ultrasonography data of a plurality of frames by the seasonal effect in time series mode; The multiresolution Knock-Down Component carries out multiresolution by level to above-mentioned ultrasonography data and decomposes, and obtains the 1st time～the n time low-frequency band decomposition view data and the 1st time～the n time high frequency band and decomposes view data, and wherein, n is the natural number more than 2; The multiresolution compound component uses above-mentioned low-frequency band to decompose view data and above-mentioned high frequency band decomposition view data execution multiresolution is synthetic, and generates composograph; Characteristic quantity is calculated parts, decomposes view data according to the above-mentioned low-frequency band that obtains and calculates characteristic quantity; The Filtering Processing parts are implemented filtering operation to the above-mentioned characteristic quantity of calculating; And high frequency band grade control part, the characteristic quantity that will decompose each corresponding time of view data utilization from above-mentioned the 1st time～the n time high frequency band that above-mentioned multiresolution Knock-Down Component obtains carries out weighting.

Description of drawings

Fig. 1 is the block diagram of the formation of expression diagnostic ultrasound equipment of the present invention.

Fig. 2 is that expression speckle of the present invention/noise reduces the figure of the detailed formation of handling part.

Fig. 3 is based on the concept map that multiresolution that the 3 D wavelet transformation that comprises time shaft carries out decomposes.

Fig. 4 is the figure that is illustrated in the use pattern of the wave filter that each change in coordinate axis direction in the 3 D wavelet transformation that comprises time shaft uses.

Fig. 5 A is used to explain the figure to the numerical differentiation of direction in space.

Fig. 5 B is used to explain the figure to the numerical differentiation of time orientation.

Fig. 6 is the flow chart of treatment step of the Filtering Processing of the Anisotropic Nonlinear diffusion filter in three times the level of expression.

Fig. 7 is the flow chart of molar behavior of the diagnostic ultrasound equipment of first embodiment.

Fig. 8 is that the speckle of accumulating and use this B mode data of the accumulating/noise that is used for explaining the B mode data of second embodiment reduces the figure that handles.

Fig. 9 is that the speckle of accumulating and use this B mode data of the accumulating/noise that is used for the B mode data of illustrative embodiment 1 reduces the figure that handles.

The specific embodiment

(first embodiment)

Below, the diagnostic ultrasound equipment of first embodiment of the present invention is described.

Fig. 1 is the block diagram of function of the diagnostic ultrasound equipment of this embodiment of expression.

As shown in Figure 1, the diagnostic ultrasound equipment of this embodiment has ultrasound probe 001, receiving and transmitting part 002, signal processing part 003, image production part 004, speckle/noise reduction handling part 100, display control unit 005 and user interface 006.And user interface 006 has display parts 007 and input parts 008 such as keyboard or mouse such as monitor.

Ultrasound probe 001 has a plurality of piezoelectric vibrators.Ultrasound probe 001 uses piezoelectric vibrator that pulse signal is transformed to ultrasound wave, and the ultrasound wave after the conversion is sent to subject.From after the receiving and transmitting part 002 input pulse signal stated.

In addition, ultrasound probe 001 utilizes piezoelectric element to receive the ultrasound wave (ultrasonic echo) that reflects from subject.And ultrasonic echo is exported to receiving and transmitting part 002 after being transformed to the signal of telecommunication (below be called " echo-signal ").To be called a frame through the set of the ultrasonic echo that obtains towards once hyperacoustic transmitting-receiving of subject from each oscillator here.That is, can generate a ultrasonography through a frame.And then, the frame scan number of times of unit interval is called frame frequency.In addition, in this embodiment, ultrasound probe 001 carries out the scanning of the same area (scanning position) of subject repeatedly under predetermined frame frequency.

In order to become the wave beam shape from the ultrasound wave boundling that ultrasound probe 001 receives, receiving and transmitting part 002 makes pulse signal arrive the time delay of each piezoelectric vibrator of ultrasound probe 001.That is, the piezoelectric vibrator at the center of ultrasound probe 001 is got over delay the time of advent of pulse signal.Through making pulse signal arrive the time delay of each piezoelectric vibrator, thereby the driving timing of each piezoelectric vibrator postpones, and the ultrasound wave boundling of generation becomes the wave beam shape.

In the reception of ultrasonic echo, also utilize 002 pair of echo-signal of receiving and transmitting part to amplify as each oscillator of analogue signal, then, make the time delay of its delay desired through (not shown) such as receive delay circuit.

Then, 002 pair of echo-signal of receiving and transmitting part is carried out the A/D conversion, then, carries out addition process based on the signal after the conversion.And then the signal that receiving and transmitting part 002 will carry out after the addition process is exported to signal processing part 003.

Signal processing part 003 has B mode treatment unit and Doppler signal processing unit.The signal that signal processing part 003 receives from receiving and transmitting part 002.

B mode treatment unit receives the signal from receiving and transmitting part 002.And then B mode treatment unit is implemented logarithmic compression and envelope detection processing etc. to received signal.In addition, B mode treatment unit generates the B mode data through the intensity apparent brightness of signal.

Doppler's processing unit carries out frequency resolution according to the signal from receiving and transmitting part 002 to velocity information.Then, Doppler's processing unit extracts the blood flow and tissue, contrast agent echo composition based on Doppler effect, and the doppler data of blood flow informations such as generation average speed, dispersion, energy.

Each data of B mode data that signal processing part 003 will be generated by B mode treatment unit and the doppler data that generated by Doppler's processing unit (below be called " initial data ") are exported to speckle/noise and are reduced handling part 100.

Ultrasound probe 001, receiving and transmitting part 002 and signal processing part 003 are equivalent to " image production part spare " among the present invention.

Speckle/noise reduces handling part 100 and is transfused to initial data from signal processing part 003.Then, speckle/noise reduces the reduction processing that 100 pairs of initial datas of being imported of handling part are carried out speckle and noise.Reduction for this speckle and noise is handled, and specifies in the back.And then speckle/noise reduces the initial data that handling part 100 will implement after the reduction processing of speckle and noise and exports to image production part 004.

Image production part 004 has DSC (Digital Scan Convertor: digital scan convertor).Image production part 004 is transfused to by speckle/noise and reduces the initial data after handling part 100 has been implemented the reduction processing of speckle and noise.

DSC is the coordinate system that is used to be presented on the monitor etc. with the initial data of being imported from the coordinate system transformation corresponding with the scanning line of ultrasonic scanning, and generates view data.

Image production part 004 has memory areas such as memorizer.In the memory area of this image production part 004, storing the view data that generates by DSC successively.These image stored data use user interface 006 to access by operators such as doctor or medical science technician (following only be called " operator ") after diagnosis.The view data that accesses is presented on the display part 007.Image production part 004 is exported to display control unit 005 with the view data that generates.

Display control unit 005 is accepted the input of view data from image production part 004.Then, display control unit 005 shows so-called B mode image and image of colorful Doppler ultrasonography like this based on the view data of input on display part 007.Below, when not distinguishing B mode image and doppler image, only be called " ultrasonography ".

The image production part 004 that has an above-mentioned functions here, is equivalent to " display control unit spare " among the present invention with display control unit 005.

Comprehensive Control portion 009 carries out the control of periodic control and the data between each function portion handing-over of the action of each function portion.Chain-dotted line among Fig. 1 is represented the flow process of control command.

(speckle/noise reduces processing)

Then, describe handling through the speckle/speckle that noise reduction handling part 100 is carried out and the reduction of noise.Speckle/noise reduction by 100 pairs of B mode datas of handling part and doppler data can both be carried out the reduction of speckle and noise and handle.Therefore, in following explanation, be treated to example with the reduction of speckle that the B mode data is carried out and noise and describe.

And, in following explanation, in order to understand easily explanation, based on the data of the ultrasonography of two dimension speckle/noise is reduced to handle describing.

In addition, the reduction that under the situation of three-dimensional ultrasonography, also can carry out speckle/noise is handled.For three-dimensional ultrasonography, be four-dimensional during factor when the joining day, but on calculating, below the calculating of the matrix of explanation only increase delegation's string.

And, in following explanation, the data that constitute a B mode image are called the B mode data of a frame.

The quantity of B mode data that in addition, will be corresponding with the number of B mode image is called frame number.The B mode data of one frame is equivalent to " the ultrasonography data " among the present invention.

Fig. 2 is that expression speckle/noise reduces the figure of the detailed formation of handling part 100.As shown in Figure 2, to carry out multiresolution synthetic in order to decompose from multiresolution, and shown in chain-dotted line, speckle/noise reduces handling part 100 and has the multiple structure that is made up of many levels (once～three time).Here, in this embodiment, in order to understand easily explanation, decomposing the synthetic high reps of multiresolution from multiresolution is three times.It is synthetic as long as get final product in once～n time (wherein, n is the natural number more than 2) scope to decompose multiresolution from multiresolution.

And, in this embodiment, decompose the use wavelet transformation as multiresolution.This wavelet transformation is the example that multiresolution decomposes, and also can use other multiresolution decomposition methods such as laplacian pyramid method or Gabor conversion.

In addition, the wavelet transformation/inverse transformation in this embodiment is meant so-called wavelet transform/inverse transformation.In following explanation, be meant that in Fig. 2 number of times more (that is, lean on more among Fig. 2 under position) is with regard to high level.

As shown in Figure 2, speckle/noise reduces handling part 100 to be possessed to each level: multiresolution decomposition portion 101 (101a, 101b, 101c); Characteristic quantity is calculated portion 102 (102a, 102b, 102c); Filtering Processing portion 103 (1031a, 103b, 103c); Multiresolution synthesizes portion 104 (104a, 104b, 104c); And high frequency band grade control part 105 (105a, 105b, 105c).

In addition, speckle/noise reduction handling part 100 possesses frame and accumulates portion 106.

Frame is accumulated portion 106 and is had storage mediums such as hard disk and memorizer.Frame is accumulated portion 106 accepts the B mode data from signal processing part 003 input.Then, frame is accumulated the B mode data of portion 106 storage from the frame that begins to obtain to the predetermined frame number (number) of up-to-date frame.The B mode data is followed effluxion and is generated successively, therefore, accumulates at frame and one after the other to accumulate the B mode data in the portion 106.This each B mode data is respectively the data of the ultrasonography that obtains of zone of the same subject of scanning.And the sweep time of the B mode data of each frame is different.Here, frame is accumulated portion 106 when being transfused to the B mode data of new frame, deletes the B mode data of the oldest frame.But frame is accumulated the storage means in the portion 106 so long as can store the method for the B mode data of necessary frame number, then also can be other method, for example can be the B mode data of not deleting old frame, the method for storing the B mode data of all frames.This frame is accumulated portion 106 and is equivalent to " data are accumulated parts before handling " among the present invention.

As shown in Figure 2, frame is accumulated portion 106 the B mode data of the predetermined frame number of being stored is exported to multiresolution decomposition portion 101.In following explanation, the set of the B mode data of predetermined frame number is called the set B mode data.

The input that portion 106 accepts the B mode data (set B mode data) of predetermined frame number is accumulated from frame by multiresolution decomposition portion 101.The multiresolution decomposition portion 101 of this embodiment implements the multiresolution decomposition for direction in space and time orientation.Here, direction in space is meant the direction in the zone of scanning subject.For example, if two-dimentional layer image then is meant the two-dimensional directional of X and Y.If the three-dimension disclocation picture then is meant the three-dimensional of X, Y and Z.In this embodiment, because ultrasonography is two-dimentional layer image, so 101 pairs in multiresolution decomposition portion comprises that the three-dimensional system of coordinate (space-time) of time shaft carries out multiresolution and decomposes.The set B mode data of 101 pairs of inputs of multiresolution decomposition portion carries out three-dimensional wavelet transformation.The set B mode data is decomposed into a kind of low-frequency band decomposes view data and seven kinds of high frequency bands decomposition view data.This multiresolution decomposition portion 101 is equivalent to " the multiresolution Knock-Down Component " among the present invention.

Fig. 3 is the concept map that decomposes through the multiresolution that the 3 D wavelet transformation that comprises time shaft carries out.

Fig. 4 is the figure that is illustrated in the use pattern of the wave filter that each change in coordinate axis direction in the 3 D wavelet transformation that comprises time shaft uses.Like Fig. 3 and shown in Figure 4; The set B mode data with space (x, y) and time (t) 300 before 101 pairs of multiresolutions of multiresolution decomposition portion decompose uses unidimensional low-frequency band (L) wave filter and high frequency band (H) wave filter respectively to each change in coordinate axis direction (each dimension) of each x, y, t rectangular axes, comes pair set B mode data 300 to carry out wavelet transformation.Through this wavelet transformation, set B mode data 300 is decomposed into a kind of low-frequency band and decomposes view data 301 and seven kind of high frequency band decomposition view data 302～308.Low-frequency band is decomposed the low-frequency component in the frequency content that view data 301 comprises that the set B mode data 300 before multiresolution decomposes had.

And each high frequency band decomposes the radio-frequency component relevant with at least one change in coordinate axis direction in the frequency content that view data 302～308 comprises that the set B mode data 300 before multiresolution decomposes had.Low-frequency band after Fig. 4 decomposes about multiresolution is decomposed view data 301 and is decomposed view data 302～308 illustrates low band filter or high band filter to each coordinate axes behaviour in service with high frequency band.Low-frequency band after multiresolution decomposes is decomposed sample number that view data 301 and high frequency band decompose view data 302～308 each comfortable each coordinate axes and is reduced into half the at the sample number of each coordinate axes of set B mode data 300 before multiresolution decomposes.

Multiresolution decomposition portion 101 makes multiresolution decompose one or many repeatedly up to arriving highest level.Thus, as shown in Figure 2, be divided into level (being 3 levels) in this embodiment.Each level shows with frequency table.Under the situation of repeatedly carrying out the multiresolution decomposition repeatedly, the low-frequency band of next level (level of a position in Fig. 2, being shown in) is decomposed the input image data of view data for decomposing with respect to next multiresolution.Particularly, do not belong in multiresolution decomposition portion 101 under the situation of highest level (in this embodiment, being three times level), the multiresolution decomposition 101a of portion decomposes view data with the low-frequency band that generates and supplies to the multiresolution decomposition 101b of portion.The multiresolution decomposition 101b of portion decomposes view data with the low-frequency band that generates and supplies to the multiresolution decomposition 101c of portion.

And the multiresolution decomposition 101c of portion that belongs to highest level (three times levels) calculates the 102c of portion with the characteristic quantity that the low-frequency band decomposition view data that generates supplies to same level (three times levels).

And the multiresolution decomposition portion 101 of each level decomposes view data with the high frequency band that generates and supplies to the high frequency band grade control part 105 that belongs to same level.

Characteristic quantity is calculated low-frequency band that portion 102 calculates supply, and to decompose the direction characteristic amount that view data comprised be marginal information.In this embodiment, calculate the information of the size at edge.Particularly, use the tangential direction at edge to calculate.Characteristic quantity is calculated the high frequency band grade control part 105 and Filtering Processing portion 103 that portion 102 arrives the information providing of the size at the edge of calculating same level.

In addition, characteristic quantity is calculated portion 102 and low-frequency band is decomposed view data is exported to Filtering Processing portion 103.This characteristic quantity is calculated portion 102 and is equivalent to " characteristic quantity is calculated parts " among the present invention.

Filtering Processing portion 103 calculates portion 102 from characteristic quantity and accepts the input of information that low-frequency band is decomposed the size at view data and edge.The low-frequency band of 103 pairs of supplies of Filtering Processing portion is decomposed view data and is used three-dimensional Anisotropic Nonlinear diffusion filter.So-called Anisotropic Nonlinear diffusion filter is meant the wave filter of stressing to carry out with nonlinear mode the composition of anisotropy diffusion.The marginal element that low-frequency band is decomposed in the view data carries out the anisotropy diffusion with nonlinear mode.Therefore, when low-frequency band was decomposed view data use Anisotropic Nonlinear diffusion filter, the brightness of marginal element increased, and non-marginal element reduces.Therefore, through using the Anisotropic Nonlinear diffusion filter, Filtering Processing portion 103 stresses the low-frequency band decomposition marginal element that view data comprised, and generates the low-frequency band that makes non-marginal element smoothed and decomposes view data.This Filtering Processing portion 103 is equivalent to " the Filtering Processing parts " among the present invention.

The Filtering Processing that characteristic quantity is calculated detection that 102 pairs of directivity characteristic quantities of portion are marginal information (in this embodiment, being the size at edge) and Filtering Processing portion 103 uses the Anisotropic Nonlinear diffusion filters to carry out is described particularly here.In this embodiment, the detection of direction characteristic amount and the Filtering Processing of using the Anisotropic Nonlinear diffusion filter are decomposed the eigenvalue and the inherent vector of the structure tensor (structure tensor) of view data and are reached through obtaining low-frequency band.The eigenvalue of this structure tensor and inherent vector represent respectively low-frequency band decompose view data the edge towards i.e. size.Especially, first eigenvalue and be illustrated in the characteristic that the low-frequency band that has planar set on the space time is decomposed view data with its corresponding respectively inherent vector.Below, suppose the low-frequency band exploded view is handled as data I, in following explanation, the low-frequency band exploded view only is called " image I " as data I.

The structure tensor of the time orientation that also comprises pixel in the image I defines as follows.

(formula 1)

$S=J_{\mathrm{\ρ}}(\▿I)=G_{\mathrm{\ρ}}*\left(\begin{array}{ccc}{I}_x^2& I_x{I}_y& I_x{I}_t\\ I_x{I}_y& I_y^2& I_y{I}_t\\ I_x{I}_t& I_y{I}_t& I_t^2\end{array}\right)=\left(\begin{array}{ccc}{s}_{11}& s_{12}& s_{13}\\ s_{12}& s_{22}& s_{23}\\ s_{13}& s_{23}& s_{33}\end{array}\right)=R\left(\begin{array}{ccc}{\mathrm{\λ}}_{s1}& 0& 0\\ 0& {\mathrm{\λ}}_{S2}& 0\\ 0& 0& {\mathrm{\λ}}_{S3}\end{array}\right)R^T$

At this, I _xThe space differentiation of the x direction of presentation video I, I _yThe space differentiation of the y direction of presentation video I, I _tThe time-derivative of the t direction (time orientation) of presentation video I, G ρ representes Gaussian function, operator " * " expression convolution.λ _S1, λ _S2, λ _S3Be the eigenvalue of structure tensor S, the magnitude relationship of its absolute value is λ _S1>=λ _S2>=λ _S3Under this situation, λ _S1It is first eigenvalue.Space differentiation I _x, I _yAnd I _tCalculation method be not limited to above-mentioned calculation method.For example, replace calculating I _x, I _yAnd I _t, the high frequency band that also can use Suo Beier wave filter (Sobel filter) or multiresolution to decompose at view data.

Ask the eigenvalue λ under the situation of each key element s of structure tensor S _S1, λ _S2, λ _S3With inherent vector R, R ^TCan calculate through known method in the linear algebra.That is the eigenvalue λ of 3 * 3 matrixes in (formula 1), _S1, λ _S2, λ _S3Method that for example can be through cardano etc. is found the solution cubic equation in fact and is calculated.And structure tensor S is a real symmetric matrix, therefore, and eigenvalue λ _S1, λ _S2, λ _S3Be real number, inherent vector R, R ^TBe real vector.Inherent vector R and R ^TMutually orthogonal.The eigenvalue λ of this structure tensor S _S1, λ _S2, λ _S3The expression edge size, inherent vector R, R ^TThe direction (tangential direction at edge) at expression edge.

In this embodiment, represent with the nonlinear diffusion equations formula (partial differential equation) shown in the following formula based on the filtering of direction characteristic.

(formula 2)

$\frac{\∂I}{\∂w}=\mathrm{div}[D\▿I]$

Here; The pixel class (brightness value) of I presentation video I;

I representes its gradient vector (gradient vector), and w representes the time factor in the diffusion equation physically.In other words, w is and handles the relevant moment, the number of processes of this diffusion equation of expression in the processing of reality.D is diffusion tensor (Diffusion Tensor), following expression.

(formula 3)

$D=\left(\begin{array}{ccc}{d}_{11}& d_{12}& d_{13}\\ d_{12}& d_{22}& d_{23}\\ d_{13}& d_{23}& d_{33}\end{array}\right)=R\left(\begin{array}{ccc}{\mathrm{\λ}}_{D1}& 0& 0\\ 0& {\mathrm{\λ}}_{D2}& 0\\ 0& 0& {\mathrm{\λ}}_{D3}\end{array}\right)R^T$

Here, R representes as following (formula 4).

(formula 4)

R＝(ω ₁ω ₂ω ₃)

Here, ω ₁, ω ₂And ω ₃Be the inherent vector of diffusion tensor D, λ _D1, λ _D2, λ _D3It is the eigenvalue of diffusion tensor D.And then, R=(ω ₁, ω ₂, ω ₃) ω ₁The normal direction at expression edge.

The eigenvalue λ of diffusion tensor D _D1Expression inherent vector ω ₁The diffusion strength of shown direction.Equally, eigenvalue λ _D2Expression inherent vector ω ₂The diffusion strength of shown direction, eigenvalue λ _D3Expression inherent vector ω ₃The diffusion strength of shown direction.And then, through controlling eigenvalue λ to each pixel _D1, λ _D2, λ _D3Value, thereby the intensity of the diffusion of control Anisotropic Nonlinear diffusion filter.In addition, the filter direction same meaning of inherent vector R and Anisotropic Nonlinear diffusion filter.That is, through suitable setting inherent vector R, thus the filtering direction of the Anisotropic Nonlinear diffusion filter of setting expectation.

Here, the inherent vector ω of diffusion tensor D ₁, ω ₂And ω ₃With the unaltered inherent vector of the structure tensor S of pixel in the above-mentioned image I or changed the vector that the order of inherent vector obtains and equate.And, the eigenvalue λ of diffusion tensor D _D1, λ _D2, λ _D3Depend on eigenvalue λ according to said structure tensor S _S1, λ _S2, λ _S3The size at the edge that calculates.

Marginal information (edge) is used the eigenvalue λ as first eigenvalue _S1With eigenvalue λ as the 3rd eigenvalue _S3Expression as following (formula 5).That is marginal information (edge) value that is to use the size at the edge in the characteristic quantity to represent.

(formula 5)

$\mathrm{edge}=1-\exp (-\frac{{({\mathrm{\λ}}_{S1}-{\mathrm{\λ}}_{S3})}^2}{k^2})$

Here, the edge have " size " and " towards ".And then, so-called " size " be the Strength Changes at image when steady for " little " edge, when the variation of the intensity of image is rapid, be the parameter at " big " edge.

Marginal information is to turn to the parameter that from 0 to 1 scope is calculated through the size criteria with the edge.The value of marginal information more near 1 just more near marginal element, more near 0 more approaching non-marginal element just.Parameter k is the parameter of the extraction degree of expression marginal element.That is,, then extract marginal element easily if reduce parameter k.

In addition, use this marginal information to calculate the eigenvalue λ of diffusion tensor D _D1, λ _D2, λ _D3Eigenvalue λ _D1, λ _D2, λ _D3Use following (formula 6), (formula 7), (formula 8) expression respectively.

(formula 6)

λ _D1＝β ₁(1-edge)+β ₂·edge

(formula 7)

λ _D2＝β ₃(1-edge)+β ₄·edge

(formula 8)

λ _D3＝β ₅(1-edge)+β ₆·edge

The part of β (1-edge) in (formula 6), (formula 7) and (formula 8) is represented non-marginal portion, and the part of β edge is represented the marginal portion.This eigenvalue λ _D1, λ _D2, λ _D3The intensity of representing diffusion as stated.And then, need make as speckle that reduces object and the non-marginal element under the noise not have directivity.That is, must make this non-marginal element independent of direction ground diffusion.Therefore, be set at β ₁=β ₃=β ₅＞0.

On the other hand, for as the marginal element that stresses object, must stress its directivity more.That is, need make vertical direction (the eigenvalue λ of this marginal element _D1Inherent vector ω ₁The direction of indication) sharpening, and make direction (eigenvalue λ in addition _D2Inherent vector ω ₂The direction of indication and eigenvalue λ _D3Inherent vector ω ₃The direction of indication) diffusion.Therefore, β ₂Be set at approximate 0 value, β ₄And β ₆Be set at and compare β ₂Big predetermined value.

As stated, can calculate the eigenvalue λ of diffusion tensor D _D1, λ _D2, λ _D3And inherent vector ω ₁, ω ₂, ω ₃Therefore, each key element d of the diffusion tensor D in (formula 3) can be calculated, (formula 2) of expression Anisotropic Nonlinear diffusion filter (diffusion equation) can be accessed.

In addition, the calculating of this Anisotropic Nonlinear diffusion filter of trying to achieve (formula 2) self is carried out through the numeric value analysis method of partial differential equation.Promptly; According to the pixel at certain some place of moment w with its around each pixel class (brightness value of pixel) and each key element value of diffusion tensor D at each point place; Obtain the new pixel class of this point at w+ Δ w place constantly; Then, w+ Δ w as new w, is carried out the same calculating of pre-determined number repeatedly.

The numerical computation method of formula 2 here, is described.Method to the numerical differentiation of time orientation at first is described.Fig. 5 A is used to explain the figure to the numerical differentiation of direction in space.Fig. 5 B is used to explain the figure to the numerical differentiation of time orientation.

The data 501 of Fig. 5 A, Fig. 5 B have unidimensional space coordinates and these two coordinates of time orientation coordinate.In Fig. 5 A, the numerical differentiation of direction in space and filtering are processing center with the data 502 of current position, and near the data 503 of the pixel of the predetermined quantity of the positive and negative both sides using are calculated.Numerical differentiation with once is an example, when use positive and negative both sides each near a pixel, the then expression of (formula 9) below the approximate usefulness of the numerical differentiation of single order.This is called 3 formula of numerical differentiation.

(formula 9)

$\frac{{\∂I}_i}{\∂x}\≈\frac{I_{i+1}-I_{i-1}}{2}$

To this, in Fig. 5 B, the numerical differentiation of time orientation and filtering are processing center with current data 504.In Fig. 5 B, can't the time in future be used for handling.For example, processing center is being set in over, in calculating formula, is using under the situation of the coefficient identical with direction in space, result can't be eliminated delay time phase in time than current frame delay.Therefore, the numerical differentiation of time orientation and filtering must be processing center with current data 504, use the data 505 of the pixel of current data 504 and predetermined quantity in the past to calculate.

Here, when function f (x) at interval [x _i, x _i+ h] consecutive hours, be expressed as

(formula 10)

$f^{\′}\left(x_i\right)=\frac{f\left(x_{i+1}\right)-f\left(x_i\right)}{h}-\frac{1}{h}\{\frac{h^2}{2!}{f}^{\′\′}\left(x_i\right)+\frac{h^3}{3!}{f}^{\′\′\′}\left(x_i\right)+...\}$

And, R _N+1Be the item that in numerical computations, becomes truncated error, if | h| is fully little, then R _N+1=0.

And then, obtain following formula through (formula 10).

(formula 11)

$f\left(x_{i-1}\right)=f\left(x_i\right)-h{f}^{\′}\left(x_i\right)+\frac{h^2}{2!}{f}^{\′\′}\left(x_i\right)-\frac{h^3}{3!}{f}^{\left(3\right)}\left(x_i\right)+...$

(formula 12)

$f\left(x_{i-2}\right)=f\left(x_i\right)-2h{f}^{\′}\left(x_i\right)+\frac{{4h}^2}{2!}{f}^{\′\′}\left(x_i\right)-\frac{{8h}^3}{3!}{f}^{\left(3\right)}+...$

And then, obtain following formula through (formula 11) and (formula 12).

(formula 13)

$f^{\′}\left(x_i\right)=\frac{3f\left(x_i\right)-4f\left(x_{i-1}\right)+f\left(x_{i-2}\right)}{2h}-\{\frac{{4h}^4}{3!}{f}^{\left(3\right)}\left(x_i\right)+...\}$

When being regarded as the error omission in { } that will be somebody's turn to do (formula 13), be expressed as

(formula 14)

$f^{\′}\left(x_i\right)\≈\frac{3f\left(x_i\right)-4f\left(x_{i-1}\right)+f\left(x_{i-2}\right)}{2h}.$

Shown in (formula 14), the approximate current data and the data of the pixel of past two frames can used of the numerical differentiation of the single order of time orientation are obtained.Be somebody's turn to do (formula 14) through using, can utilize the numerical computation method of (formula 2) to calculate.

Particularly, (formula 2) can be expressed as following (formula 15).

(formula 15)

$\frac{\∂I}{\∂w}=\mathrm{div}[D\▿I]=\mathrm{div}\left[\left(\begin{array}{ccc}{d}_{11}& d_{12}& d_{13}\\ d_{12}& d_{22}& d_{23}\\ d_{13}& d_{23}& d_{33}\end{array}\right){\left(\begin{array}{ccc}\frac{\∂I}{\∂x}& \frac{\∂I}{\∂y}& \frac{\∂I}{\∂t}\end{array}\right)}^T\right]$

$=\frac{\∂}{\∂x}\left(d_{11}\frac{\∂I}{\∂x}\right)+\frac{\∂}{\∂x}\left(d_{12}\frac{\∂I}{\∂y}\right)+\frac{\∂}{\∂x}\left(d_{13}\frac{\∂I}{\∂t}\right)$

$+\frac{\∂}{\∂y}\left(d_{12}\frac{\∂I}{\∂x}\right)+\frac{\∂}{\∂y}\left(d_{22}\frac{\∂I}{\∂y}\right)+\frac{\∂}{\∂y}\left(d_{23}\frac{\∂I}{\∂t}\right)$

$+\frac{\∂}{\∂t}\left(d_{13}\frac{\∂I}{\∂x}\right)+\frac{\∂}{\∂t}\left(d_{23}\frac{\∂I}{\∂y}\right)+\frac{\∂}{\∂z}\left(d_{33}\frac{\∂I}{\∂t}\right)$

For the partial differential of each item of calculating this (formula 15), use the approximation of above-mentioned (formula 9) and (formula 14).

The treatment step of above Anisotropic Nonlinear diffusion filter here, is described with reference to Fig. 6.Fig. 6 is the flow chart of treatment step of the Filtering Processing of the Anisotropic Nonlinear diffusion filter in three times the level of expression.Here, be that example describes with the treatment step of the Filtering Processing of three times levels, but once or also carry out same processing in the Filtering Processing of secondary.

Step S001: the multiresolution decomposition 101c of portion carries out the input of the data of Filtering Processing from the multiresolution decomposition 101b of portion acceptance.

Step S002: the multiresolution decomposition 101c of portion to the input data at each change in coordinate axis direction (I _x, I _y, I _t) differential and generate low-frequency band and decompose view data and high frequency band decomposition view data.

Step S003: characteristic quantity is calculated the 102c of portion each key element of decomposing the structure tensor S of view data from the low-frequency band of the multiresolution decomposition 101c of portion input is calculated.In this calculates, also comprise the calculating of the Gaussian function G ρ that computation structure tensor S is preceding.

Step S004: characteristic quantity is calculated the 101c of portion and is calculated its eigenvalue λ according to structure tensor S _S1, λ _S2, λ _S3And inherent vector ω ₁, ω ₂, ω ₃

Step S005: characteristic quantity is calculated the eigenvalue λ of the 102c of portion according to structure tensor S _S1, λ _S2, λ _S3Obtain the size at edge, according to inherent vector ω ₁, ω ₂, ω ₃Obtain the edge towards.

Step S006: characteristic quantity is calculated the high frequency band grade control part 105c that the 102c of portion exports to the size at the edge of obtaining same level.

Step S007: the 103c of Filtering Processing portion calculates the intensity of diffusion according to the size of being calculated the edge that the 102c of portion obtains by characteristic quantity, in addition, and according to the intensity of the diffusion of obtaining and the inherent vector ω of structure tensor S ₁, ω ₂, ω ₃Calculate each coefficient that uses in the numeric value analysis of diffusion equation (formula 2) at the Anisotropic Nonlinear diffusion filter.

Step S008: the numeric value analysis that the 103c of Filtering Processing portion carries out diffusion equation calculates.

Step S009: the numeric value analysis that the 103c of Filtering Processing portion judges whether to have carried out the diffusion equation of pre-determined number calculates.Do not carrying out turning back to step S008 under the situation of pre-determined number.Under the situation of having carried out pre-determined number, get into step S010.

Step S010: the 103c of Filtering Processing portion will carry out the low-frequency band after the Filtering Processing through the Anisotropic Nonlinear diffusion filter and decompose the synthetic 104c of portion of multiresolution that view data is exported to same level.

High frequency band grade control part 105 is accepted the input that eight high frequency bands decompose view data from the multiresolution decomposition portion 101 of same level.In addition, high frequency band grade control part 105 is calculated the input that portion 102 accepts the size at edge from characteristic quantity.The information (size at edge) at high frequency band grade control part 105 use edges is decomposed the processing that view data is stressed the marginal portion to high frequency band.In this embodiment, the size that high frequency band grade control part 105 is got the edge that the eigenvalue by diffusion tensor D calculates and each high frequency band decompose each pixel long-pending of view data.In addition, this long-pending result multiply by the control coefrficient that each high frequency band decomposes view data.Through getting long-pending with the size at edge like this, can stress that high frequency band decomposes the marginal portion in the view data, can suppress the part beyond the edge.

As processing method of data, has the method that the control coefrficient of each high frequency imaging is multiply by in zone beyond the edge as the high frequency band exploded view of the size (marginal information) at the edge that uses other.Through carrying out this processing, in high frequency band decomposition view data, also can stress the marginal portion.The high frequency band that high frequency band grade control part 105 will have been implemented the processing of emphasical above-mentioned marginal portion decomposes the synthetic portion 104 of multiresolution that view data is exported to same level.This high frequency band grade control part 105 is equivalent to " the high frequency band grade control assembly " among the present invention.

The synthetic portion 104 of multiresolution accepts the input that a low-frequency band is decomposed view data from Filtering Processing portion 103.

And the synthetic portion 104 of multiresolution accepts the input that seven high frequency bands decompose view data from high frequency band grade control part 105.A low-frequency band of 104 pairs of inputs of the synthetic portion of multiresolution is decomposed view data and seven high frequency bands and is decomposed view data and synthesize and generate a composograph.When generating composograph, the synthetic portion 104 of multiresolution decomposes view data according to a low-frequency band and seven high frequency bands decomposition view data are carried out wavelet inverse transformation.The composograph of this generation is two times from the image of Filtering Processing portion 103 inputs at the sample number of each coordinate axes.The synthetic portion 104 of multiresolution in the level more than the secondary (in this embodiment, is synthetic 104b of portion of multiresolution or 104c.) characteristic quantity that formed image is exported to next level calculates portion 102.

And the multiresolution in the level once synthesizes portion 104 and (in this embodiment, is the synthetic 104a of portion of multiresolution.) formed image is exported to image production part 004.The synthetic portion 104 of this multiresolution is equivalent to " multiresolution compound component " among the present invention.

Here, with reference to Fig. 7, the flow process of the molar behavior of the diagnostic ultrasound equipment of this embodiment is described.Fig. 7 is the flow chart of molar behavior of the diagnostic ultrasound equipment of this embodiment.Here, be that example describes also with the B mode image.

Step S101: receiving and transmitting part 002 sends ultrasound wave via 001 pair of subject of ultrasound probe, and receives the echo-signal of the ultrasonic echo that reflects based on this subject.

Step S102: 003 pair of echo-signal from receiving and transmitting part 002 input of signal processing part is carried out signal processing such as logarithmic compression and envelope detection.

Step S103: speckle/noise reduction handling part 100 will be stored in frame to each frame from the B mode data of signal processing part 003 input and accumulate portion 106.

Step S104: speckle/noise reduces handling part 100 and accumulates the B mode data of up-to-date B mode data before the predetermined frame number in the portion 106 according to being stored in frame; Carry out that multiresolution decomposes, characteristic quantity is calculated, Filtering Processing, the control of high frequency band grade and multiresolution be synthetic, and up-to-date B mode data is carried out speckle/noise reduces and handle.

Step S105: 004 pair of image production part reduce from speckle/noise handling part 100 inputs carrying out speckle/noise reduce the B mode data of handling and carry out coordinate axis transform etc., and generate the ultrasonography of B mode data.

Step S106: display control unit 005 will be shown in display part 007 from the ultrasonography of image production part 004 input.

Step S107: Comprehensive Control portion 009 judges whether inspection finishes.Under the unclosed situation of inspection, turn back to step S101.Under the situation that inspection is through with, stop the action of diagnostic ultrasound equipment.

More than, reduce the object of handling as speckle/noise and use two-dimentional layer image to be illustrated.Reduce the object of handling as speckle/noise and also can use three-dimension disclocation pictures such as three-dimensional volume data.Reduce processing in order three-dimension disclocation to be looked like carry out speckle/noise, as long as make above-mentioned processing increase one dimension.Particularly, as long as calculating of on the edge of size makes matrix increase delegation's string in the calculating with Filtering Processing to calculate.

More than, speckle/noise reduces handling part 100 and is disposed at after the signal processing part 003.Speckle/noise reduces handling part 100 and also can be configured in after the image production part 004.Under the situation after speckle/noise reduction handling part 100 is configured in image production part 004, the ultrasonography that has carried out coordinate axis transform is carried out speckle/noise reduce processing.In this case, ultrasound probe 001, receiving and transmitting part 002, signal processing part 003 and image production part 004 are equivalent to " image production part spare " among the present invention, and display control unit 005 is equivalent to " the display control unit spare " among the present invention.

That kind as described above, the direction in space of the initial data of the diagnostic ultrasound equipment of this embodiment in each pattern of B pattern or doppler mode and time orientation are carried out the multiresolution decomposition, calculate the direction characteristic amount.Utilize the wave filter of this direction characteristic amount through use, thereby reduced direction in space and the noise of time orientation and the variation of speckle.Thus, not only at direction in space, on time orientation, also can generate the image of level and smooth and accentuate tissue composition.

And the diagnostic ultrasound equipment of this embodiment is such formation: about the numerical differentiation to time orientation, use data current and a plurality of frames in the past, calculate through formula different with direction in space and coefficient.Thus, can the have no time Filtering Processing of Phase delay.

And, in this embodiment, use 105 pairs of high frequencies of high frequency band grade control part to decompose view data and handle, stress that high frequency band decomposes the marginal portion of view data, thus, generate the image of further stressing the edge.

But, also can not use high frequency band grade control part 105, high frequency band is decomposed view data do not handle, directly use high frequency band to decompose view data at that, in the synthetic portion 104 of multiresolution, generate composograph.In this case, compare with the situation of using high frequency band grade control part 105, the emphasical degree of component of organization weakens, but can generate level and smooth image to time orientation.

In addition, multiresolution decomposition portion 101, characteristic quantity calculate synthetic portion 104 of portion 102, Filtering Processing portion 103, multiresolution and high frequency band grade control part 105 also can be directed against each level (among Fig. 2 once～n time at every turn) be configured.Also can calculate the processing that portion 102, Filtering Processing portion 103, the synthetic portion 104 of multiresolution and high frequency band grade control part 105 carry out each level (each time) successively through one group multiresolution decomposition portion 101, characteristic quantity.

[second embodiment]

Below, the diagnostic ultrasound equipment of second embodiment of the present invention is described.The diagnostic ultrasound equipment of this embodiment and the first embodiment difference are, are to utilize to have carried out the B mode data that speckle/noise reduction is handled carries out speckle/noise reduction processing to next frame formation.Utilize the block diagram of Fig. 1 that the formation of the diagnostic ultrasound equipment of this embodiment is shown.In following explanation, added the short of special instruction of function portion of the label identical with first embodiment, then have identical functions.

Fig. 8 is the figure of the processing of accumulating and use this B mode data of accumulating that is used for explaining the B mode data of this embodiment.In order to understand explanation easily, Fig. 8 has explained spatial axes and this two dimension of time shaft.

Frame accumulate portion 106 to each time through storing successively from the B mode data of signal processing part 003 input.Here, with up-to-date B mode data as B mode data 801.And then speckle/noise reduces handling part 100 and uses 300 pairs of B mode datas of set B mode data 801 to carry out speckle/noise reduction processing.Set B mode data 300 is to comprise being stored in the set of B mode data that frame is accumulated the predetermined frame number of the B mode data 801 in the portion 106.

And frame is accumulated the B mode data 801 before portion 106 deletion is handled, and will carry out the B mode data that the B mode data of handling 801 (below, be called and handle back B mode data 801) is stored as up-to-date frame.Then, when from the B mode data (being referred to as B mode data 802) of signal processing part 003 input next frame, B mode data 802 is stored in frame as up-to-date frame accumulates in the portion 106.Then, speckle/noise reduces by 100 pairs of B mode datas of handling part 802 and carries out speckle/noise and reduce and handle, and frame is accumulated B mode data 802 after portion's 106 stores processor.Through carrying out the processing of above that kind repeatedly, thus frame accumulate portion 106 and can accumulate the result of B mode data in the past at time orientation, and the speckle/noise that uses B mode data after this processing of accumulating to carry out up-to-date B mode data reduces and handles.Here, frame is accumulated portion 106 and is equivalent to " handle back frame accumulate parts " among the present invention.

Like this, the diagnostic ultrasound equipment of this embodiment is such formation: the speckle/noise that uses the B mode data of having implemented speckle/past that noise reduction is handled to carry out up-to-date B mode data reduces to be handled.

Thus, can use speckle/noise to reduce the level and smooth B mode data of having implemented processing in the processing, therefore, can further reduce the variation of the speckle/noise of time orientation effectively, the processing of the Phase delay that has no time at time orientation.

(variation 1)

Then, other example to diagnostic ultrasound equipment of the present invention describes.In this variation, accumulate the B mode data before handling.Reduce when handling carrying out speckle/noise, the frame before the up-to-date processing is exported to frame accumulate portion 106.

Fig. 9 is the figure of the processing of accumulating and use this B mode data of accumulating that is used for explaining the B mode data of present embodiment.Fig. 9 representes with spatial axes and this two dimension of time shaft.

Diagnostic ultrasound equipment of the present invention is following formation: frame is accumulated portion 107 before in the diagnostic ultrasound equipment of second embodiment, further having appended the processing that dots.

The preceding frame of processing is accumulated portion 107 and is had storage mediums such as memorizer and hard disk.

And, before handling frame accumulate portion 107 and time through storing successively accordingly from the B mode data of signal processing part 003 input.Frame is accumulated the B mode data that portion 107 can store all frames before handling, and also can store the B mode data of predetermined frame number.

In the formation of the B mode data of storing predetermined frame number, when the B mode data 901 before the new processing of input, delete the preceding B mode data of the oldest processing.

Frame was accumulated the input that portion 107 requires the preceding B mode data 901 of speckle/noise reduction processing before frame was accumulated 106 pairs of processing of portion.Frame is accumulated portion 107 and is accumulated the B mode data 901 before portion 106 imports up-to-date processing to frame accordingly with this input requirement before handling.

Frame accumulated carrying out before portion's 106 storing predetermined number-1 frames speckle/noise reduce the B mode data of handling 902.Frame accumulate portion 106 will be before handle the B mode data 902 accumulated after the processing of B mode data 901 and storage before the processing that portion 107 duplicates of frame export to multiresolution decomposition portion 101.Then, frame is accumulated the B mode data the processing of speckle/noise reduction processing has been carried out in portion's 106 storages through speckle/noise reduction handling part 100 after.At this moment, frame is accumulated B mode data after the oldest processing of portion 106 deletion.

Through such formation, can also further reduce the variation of the speckle/noise of time orientation effectively, the processing of the Phase delay that has no time.

Claims (8)

1. a diagnostic ultrasound equipment is characterized in that, this diagnostic ultrasound equipment possesses:

Image production part spare, through the execution ultrasonic transmission/reception, thereby to generate the ultrasonography data of a plurality of frames by the seasonal effect in time series mode;

The multiresolution Knock-Down Component; By level above-mentioned ultrasonography data are carried out the multiresolution decomposition for direction in space and time orientation; Obtain the 1st time～the n time low-frequency band decomposition view data and the 1st time～the n time high frequency band and decompose view data, wherein, n is the natural number more than 2;

Characteristic quantity is calculated parts, decomposes view data according to the above-mentioned low-frequency band that obtains and calculates characteristic quantity;

The Filtering Processing parts according to the above-mentioned characteristic quantity of calculating, decompose view data to above-mentioned low-frequency band and apply the Anisotropic Nonlinear diffusion filter; And

The multiresolution compound component, it is synthetic to use above-mentioned low-frequency band decomposition view data and above-mentioned high frequency band decomposition view data to carry out multiresolution, thereby generates composograph.

2. diagnostic ultrasound equipment according to claim 1 is characterized in that,

This diagnostic ultrasound equipment also possesses the data of the above-mentioned ultrasonography data of storage and accumulates parts,

As the Filtering Processing in the ultrasonography data of the t frame in above-mentioned a plurality of frames; Above-mentioned Filtering Processing parts use be stored in above-mentioned data accumulate in the parts from the t-1 frame to predetermined frame before the ultrasonography data of each frame each pixel value of each pixel of same position diffusion equation is carried out numeric value analysis, calculate the pixel value of each pixel of above-mentioned same position of the ultrasonography data of t frame.

3. diagnostic ultrasound equipment according to claim 1 is characterized in that,

This diagnostic ultrasound equipment also possesses the data of the above-mentioned low-frequency band decomposition of storage view data and accumulates parts,

As the Filtering Processing in the ultrasonography data of t frame; Above-mentioned Filtering Processing parts use be stored in above-mentioned data accumulate in the parts from the t-1 frame to predetermined frame before the above-mentioned low-frequency band of each frame each pixel value of decomposing each pixel of the same position the view data diffusion equation is carried out numeric value analysis, calculate the pixel value of each pixel that low-frequency band after the processing of t frame is decomposed the above-mentioned same position of view data.

4. diagnostic ultrasound equipment according to claim 1 is characterized in that,

Above-mentioned Filtering Processing parts use the wave filter of being represented by the nonlinear diffusion equations formula,

Under the situation of the numeric value analysis that carries out this nonlinear diffusion equations formula,

As the numerical differentiation of direction in space, use

$\∂I_i/\∂x=\{I_{i+1}-I_{i-1}\}/2,$

As the numerical differentiation of time orientation, use

$\∂I_t/\∂t=\{3I_t-4I_{t-1}-I_{t-2}\}/2,$

I _i: the pixel value when x is i, I _t: the pixel value of moment t.

5. diagnostic ultrasound equipment according to claim 1 is characterized in that,

Above-mentioned characteristic quantity is calculated the structure tensor that parts are calculated each pixel, and uses the said structure tensor to calculate eigenvalue and inherent vector, uses the above-mentioned eigenvalue and the inherent vector of calculating characteristic quantity.

6. diagnostic ultrasound equipment according to claim 1 is characterized in that,

Above-mentioned Filtering Processing parts are the Anisotropic Nonlinear diffusion filter.

7. diagnostic ultrasound equipment according to claim 1 is characterized in that,

This diagnostic ultrasound equipment also possesses high frequency band grade control assembly; This high frequency band grade control assembly decomposes the Filtering Processing that view data is stressed the edge to above-mentioned high frequency band, and the decomposition of the high frequency band after will handling view data is exported to above-mentioned multiresolution compound component.

8. a diagnostic ultrasound equipment is characterized in that, this diagnostic ultrasound equipment possesses:

Image production part spare, through the execution ultrasonic transmission/reception, thereby to generate the ultrasonography data of a plurality of frames by the seasonal effect in time series mode;

The multiresolution Knock-Down Component; By level above-mentioned ultrasonography data are carried out the multiresolution decomposition for direction in space and time orientation; Obtain the 1st time～the n time low-frequency band decomposition view data and the 1st time～the n time high frequency band and decompose view data, wherein, n is the natural number more than 2;

The multiresolution compound component uses above-mentioned low-frequency band to decompose view data and above-mentioned high frequency band decomposition view data execution multiresolution is synthetic, and generates composograph;

Characteristic quantity is calculated parts, decomposes view data according to the above-mentioned low-frequency band that obtains and calculates characteristic quantity;

The Filtering Processing parts according to the above-mentioned characteristic quantity of calculating, decompose view data to above-mentioned low-frequency band and apply the Anisotropic Nonlinear diffusion filter; And

High frequency band grade control part, the characteristic quantity that will decompose each corresponding time of view data utilization from above-mentioned the 1st time～the n time high frequency band that above-mentioned multiresolution Knock-Down Component obtains carries out weighting.

Images